Quality Adjustment, Hedonic Regressions and the Extension Problem

High technology products are characterized by the rapid introduction of new models and the corresponding disappearance of older models. The paper addresses the quality adjustment problem associated with the construction of price indexes for these products. A main method for dealing with this problem is the use of hedonic regression models. Hedonic regressions use either product characteristics as explanatory variables (Time Dummy Characteristics regressions) or the product itself as the ultimate characteristic (Time Product Dummy regressions). The paper considers weighted and unweighted Time Product Dummy regressions. The indexes which were generated by the hedonic regressions are compared to traditional index numbers that did not make any special adjustments for quality change. The Expanding Window variant of a Weighted Time Product Dummy regression was used to address the chain drift problem and the problems associated with extending a series that cannot be revised. Finally, the estimation of systems of inverse demand functions was also used to generate various price indexes. Seventeen alternative approaches were implemented using Japanese price and quantity data on laptop sales in Japan for the 24 months over the years 2020-2021.